• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.11a
• /
• pp.705-708
• /
• 2023

다중밀집 사고를 사전에 방지하기 위해 군중 밀집도를 정확하게 파악하는 것은 중요하다. 기존 방법 중 일부는 군중 계수를 기반으로 군중 밀집도를 추정하거나 원근 왜곡이 있는 데이터를 그대로 학습한다. 이 방식은 물체의 거리에 따라 크기가 달라지는 원근 왜곡에 큰 영향을 받는다. 본 연구는 보행자 깊이 정보를 이용한 군중 밀집도 알고리즘을 제안한다. 보행자의 깊이 정보를 계산하기 위해 편차가 적은 머리 크기를 이용한다. 머리를 탐지하기 위해 OC-Sort를 학습모델로 사용한다. 탐지된 머리의 경계박스 좌표, 실제 머리 크기, 카메라 파라미터 등을 이용하여 보행자의 깊이 정보를 추정한다. 이후 깊이 정보를 기반으로 밀도 맵을 추정한다. 제안 알고리즘은 혼잡한 환경에서 객체의 위치와 밀집도를 정확하게 분석하여 군중밀집 사고를 사전에 방지하는 지능형 CCTV시스템의 기반 기술로 활용될 수 있으며, 더불어 보안 및 교통 관리 시스템의 효율성을 향상하는 데 중요한 역할을 할 것으로 기대한다.

• Korean Journal of Optics and Photonics
• /
• v.14 no.4
• /
• pp.443-448
• /
• 2003

A study on optimum design of the lightweight mirror of a satellite camera is presented. An optical surface deformation of the lightweight mirror, which is a principal component of the camera system, is an important factor affecting the optical performance of the whole camera system. In this study, optimum design of the lightweight mirror is presented. Total weight of the mirror to reduce the optical surface deformation and the launching cost is used as an objective function. Peak-to-valley value and natural frequency of the mirror are given as constraints to the optimization problem. The sensitivities of the objective function and constraint are calculated by a finite difference method. The optimization procedure is carried out by the commercial optimizer, DOT. As a verification of the optimum design of the mirror, two design examples are treated. In the real application example, the lightweight mirror with 600mm effective diameter is treated. The optimized results with various design variables, which are obtained by considering thickness limitations, are analyzed.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2007.02a
• /
• pp.3-8
• /
• 2007

최근, 실감 영상에 대한 관심과 요구가 증가하면서 신개념 서비스인 3차원 다시점(Multi-view) 방송에 대한 연구가 다양하게 진행되고 있다. 이와 더불어 광고와 게시를 목적으로 입체 영상과 입체 디스플레이 장치의 수요가 증가하고 있어, 앞으로 다시점 영상 콘텐츠와 디스플레이 장치가 활발하게 보급될 전망이다. 다시점 영상 콘텐츠는 제작 단계에서 컴퓨터 그래픽스 객체를 합성하면 보다 목적에 부합하는 콘텐츠를 제작할 수 있다. 본 논문에서는 다시점 카메라로부터 얻은 RGB 텍스쳐 데이터와 깊이 테이터에 컴퓨터 그래픽스 객체를 합성하여 다시점 합성 영상을 생성하는 방법을 제안한다. 또한, 제작된 다시점 합성 영상을 검증하고 재생하는 다시점 비디오 뷰어를 설계, 구현 한다. 가상의 다시점 영상에 그래픽스 객체를 합성하는 방법은 후 합성 기반으로, 임의의 그래픽스 객체 모델을 생성하여 깊이 정보를 부여하고, 가상 시점 영상의 생성과 동일한 방법으로 그래픽스 객체의 각 시점별 영상을 생성한다. 끝으로 깊이정보를 사용하여 가상 시점 영상의 적절한 좌표공간으로 그래픽스 객체를 삽입한다. 그래픽스 합성의 정확성 검증을 위해 다시점 그래픽스 합성 영상을 디스플레이하는 뷰어는 2D 및 입체를 모두 지원하고, view switching, frozen moment, view sweeping 등의 interactive special effect기법과 다양한 포맷의 저장이 가능하다. 또한, 입체 영상의 실험에서는 그래픽 객체의 입체감 조절을 위해 실제 카메라 시점 간에 필요한 중간시점영상의 개수를 결정할 수 있다.

• Journal of KIISE
• /
• v.41 no.11
• /
• pp.900-910
• /
• 2014

Multi-view video is obtained by capturing one three-dimensional scene with many cameras at different positions. The computational complexity of multi view video coding increases in proportion to the number of cameras. To reduce computational complexity and maintain the image quality, improved prediction structure and motion estimation method is proposed in this paper. The proposed prediction structure exploits an average distance between the current picture and its reference pictures. The proposed prediction structure divides every GOP into several groups to decide the maximum index of hierarchical B layer and the number of pictures of each B layer. And the proposed motion estimation method uses a hierarchical search strategy. This strategy method consists of modified diamond search pattern, progressive diamond search pattern and modified raster search pattern. Experiment results show that the complexity reduction of the proposed prediction structure and motion estimation method over JMVC (Joint Multiview Video Coding) reference model using hierarchical B pictures of Fraunhofer-HHI and TZ search method can be up to 40~70% while maintaining similar video quality and bit rates.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.15 no.7
• /
• pp.1564-1572
• /
• 2011

In this paper, we propose a new histogram based hand recognition algorithm for augmented reality. Hand recognition system makes it possible a useful interaction between an user and computer. However, there is difficulty in vision-based hand gesture recognition with viewing angle dependency due to the complexity of human hand shape. A new hand recognition system proposed in this paper is based on the features from hand geometry. The proposed recognition system consists of two steps. In the first step, hand region is extracted from the image captured by a camera and then hand gestures are recognized in the second step. At first, we extract hand region by deleting background and using skin color information. Then we recognize hand shape by determining hand feature point using histogram of the obtained hand region. Finally, we design a augmented reality system by controlling a 3D object with the recognized hand gesture. Experimental results show that the proposed algorithm gives more than 91% accuracy for the hand recognition with less computational power.

• Journal of Internet Computing and Services
• /
• v.7 no.3
• /
• pp.107-118
• /
• 2006

Recently, network infra grows rapidly and the digital image compression technique have made remarkable progress, and therefore the demand of the real-time image surveillance system which uses a network camera server is increasing. Network Camera Server has emerged as an attractive alternative to the CCTV for the real-time image surveillance. From this reason, the demand regarding the development of the real-time image surveillance system which uses a network camera server is increasing as well. In this paper, we will provide a model for JPEG Streamer which is used in real-time image surveillance system. And we will design and implement this model. For the stability and efficiency of the JPEG Streamer, we'll use the thread pool and shared memory. We will also introduce the concept of double buffering for increasing the quality of real-time image, Using the JPEG Streamer, a raising of the productivity, reliability and stability will be able to secure to development of real-time image surveillance system.

• The Journal of the Korea Contents Association
• /
• v.10 no.8
• /
• pp.11-17
• /
• 2010

In this paper, we propose HMM-based upper-body gesture. First, to recognize gesture of space, division about pose that is composing gesture once should be put priority. In order to divide poses which using interface, we used two IR cameras established on front side and side. So we can divide and acquire in front side pose and side pose about one pose in each IR camera. We divided the acquired IR pose image using SVM's non-linear RBF kernel function. If we use RBF kernel, we can divide misclassification between non-linear classification poses. Like this, sequences of divided poses is recognized by gesture using HMM's state transition matrix. The recognized gesture can apply to existent application to do mapping to OS Value.

• Journal of the Korean Society of Marine Environment & Safety
• /
• v.26 no.6
• /
• pp.601-605
• /
• 2020

In large ships with complex structures, it is difficult to locate workers. In particular, it is not easy to detect when a worker falls down, making it difficult to respond quickly. Thus, research is being conducted to detect fallen workers using a camera or by attaching a device to the body. Existing image-based fall detection systems have been designed to detect a person's body parts; hence, it is difficult to detect them in various ships and postures. In this study, the entire fall area was extracted and deep learning was used to detect the fallen shipworker based on the image. The data necessary for learning were obtained by recording falling states at the shipyard. The amount of learning data was augmented by flipping, resizing, and rotating the image. Performance evaluation was conducted with precision, reproducibility, accuracy, and a low error rate. The larger the amount of data, the better the precision. In the future, reinforcing various data is expected to improve the effectiveness of camera-based fall detection models, and thus improve safety.

• The Journal of the Korea Contents Association
• /
• v.10 no.10
• /
• pp.120-127
• /
• 2010

Today we always can take a picture and enjoy it anywhere and anytime with a propagation of digital camera and camera phone. Even though the amount of digital picture content produced in every day life has been increasing, the way of enjoying it was just with phones or PC monitors. A small size screen of a cell phone and a desktop monitor are not suitable environment for enjoying digital picture, so the new market for digital photo frame has been opened. Until now, various kind of digital photo frames of many manufactures have been released for several years, but they have similar interaction way each other due to the limitation of their hardware specification so don't fill the intuitive mental model that users have about the product called a photo frame. Thus, this research aims to suggest an interaction way of digital photo frame providing with better usability and emotional satisfaction according to the spatial environment that it's located and users' behaviors.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.50 no.1
• /
• pp.225-231
• /
• 2013

In this study, we propose the design of optimized pRBFNNs-based night vision face recognition system using PCA algorithm. It is difficalt to obtain images using CCD camera due to low brightness under surround condition without lighting. The quality of the images distorted by low illuminance is improved by using night vision camera and histogram equalization. Ada-Boost algorithm also is used for the detection of face image between face and non-face image area. The dimension of the obtained image data is reduced to low dimension using PCA method. Also we introduce the pRBFNNs as recognition module. The proposed pRBFNNs consists of three functional modules such as the condition part, the conclusion part, and the inference part. In the condition part of fuzzy rules, input space is partitioned by using Fuzzy C-Means clustering. In the conclusion part of rules, the connection weights of pRBFNNs is represented as three kinds of polynomials such as linear, quadratic, and modified quadratic. The essential design parameters of the networks are optimized by means of Differential Evolution.